Translated site not selling? Here's what you need to change

Companies expanding internationally usually start the same way: translate the site, launch ads, wait for results. When results don't come, the conclusion is always the same: "this market doesn't work." But the problem is rarely the market. It's how the site was prepared.

A translated site is not the same as a site ready to sell. Localization covers far more than swapping text from one language to another. It's adapting the entire experience: how Google sees your pages, what checkout looks like, what language the order confirmation arrives in.

This guide walks through every element that needs to work for a website or online store to actually sell in a new market. Step by step, with concrete examples.

URL structure and technical foundations

Before you translate anything, you need your technical foundations in order. Without them, Google won't properly index your language versions, no matter how good the content is.

First decision: subdirectories (/de/, /cs/, /fr/) or subdomains (de.yourdomain.com). Both approaches work, but pick one and stick with it. Mixing structures is asking for indexation problems.

What you need in place:

• A separate XML sitemap for each language version. One shared sitemap for 6 languages is a mess that slows down Google's discovery of new pages.
• Separate Google Search Console verification for each version. Without this, you can't see indexation data, errors, or keyword performance per market. You're flying blind.
• Correct canonical tags. A canonical from /de/ pointing to /pl/ will de-index the German version. Each language version must self-reference.
• 301 redirects for old URLs. If you previously had a different structure, incomplete redirects are one of the most common reasons Google serves the wrong language version.

This is the foundation of multilingual technical SEO. Without it, the next steps are pointless because Google doesn't know what to show to whom.

Hreflang. How not to break it

Hreflang tells Google: "this page has a German version at this URL, a Czech one here, and a Polish one there." Sounds simple. In practice, it's one of the most commonly botched implementations in multilingual sites.

Rules that must be met:

• Reciprocity. If /pl/ points to /de/, then /de/ must point back to /pl/. With 6 language versions, that's 6 hreflang tags on every URL. No reciprocity = Google ignores the tag.
• x-default. The fallback version for users whose language you don't support. Without x-default, Google guesses what to show. It usually guesses wrong.
• Correct codes. "en" is not the same as "en-GB." If you're targeting the UK market, you need en-GB. If you're targeting English speakers generally, en is fine. Wrong codes are a common issue, especially with country-specific versions.
• No conflict with canonicals. A canonical from /de/ pointing to /pl/ while hreflang says "/de/ is the German version" sends contradictory signals. Google doesn't know which to trust.

In one marketplace project, simply fixing the hreflang implementation, with no content changes and no new pages, tripled organic traffic on the international versions. That's how much potential was being blocked by technical debt.

Keywords. Research from scratch, not translation

This is the trap most companies fall into: take your source-language keywords, translate them into German or Czech, and plug them into your content. The problem is that translated keywords often hit queries with zero volume or a completely different intent.

A concrete example: "amusement park for kids" in Czech is "zábavní park pro děti." That phrase has strong informational volume in Czechia. Its English equivalent barely registers in the same way. Without local research, it would never enter the content strategy for the Czech market.

Or e-commerce: "women's running shoes" in German. "Damen Laufschuhe" and "Laufschuhe für Damen" mean the same thing but have very different search volumes. Choosing the wrong variant means less traffic with the same content quality.

What you need to do:

• Run keyword research separately for each market. Use tools (Ahrefs, Semrush) set to the target country. Don't translate from your source keyword list.
• Map search intent. The same phrase can be informational in one country and transactional in another. Your content strategy has to reflect that.
• Create content from scratch, don't translate it. WordPress content localization or any other CMS requires a separate strategy per market, not a copy-paste from the original version.

CMS and platform. What gives you control vs. what limits you

Not every e-commerce platform and not every CMS handles multilingual the same way. Your technology choice affects how much work localization requires and where you'll hit limitations.

Shopify with Shopify Markets lets you manage multiple markets from one dashboard. Multilingual support is solid but requires a translation app (Translate & Adapt, Langify) and manual hreflang configuration. Shopify store localization on the Plus plan gives full checkout control. On standard plans, that control is limited.

WooCommerce with WPML or Polylang offers great flexibility but demands more technical work. Each language version is a separate set of pages, which means more maintenance. The upside is full code control. WooCommerce store translation scales well across multiple markets, provided someone on your team knows their way around WPML.

Magento (Adobe Commerce) has built-in multilingual support via Store Views. It's an enterprise solution that scales well with many markets but requires an experienced technical team.

PrestaShop and Shoper have basic multilingual support. Fine for one or two markets. At five markets and thousands of products, the limitations become noticeable.

Regardless of platform, the key question is: do you have control over checkout, hreflang, and transactional emails? If the platform makes that difficult, localization will be more expensive and slower.

Checkout. This is where you lose or make money

You can have great Google rankings and well-performing ads, but if the customer reaches checkout and something feels off, you lose them at the last step. And in international markets, that friction comes much faster than it does at home.

Baymard Institute data: the average cart abandonment rate is 70%. In markets where the customer isn't comfortable with the language, currency, or payment method, that number is even higher.

What needs to be localized:

• Payment methods the customer knows. Germans buy with Klarna and SEPA. The Dutch use iDEAL. Belgians use Bancontact. Missing the preferred payment method is the single most common reason for cart abandonment in international markets.
• Address form format. Field order, postal code format, required fields. A form that looks "off" instantly raises suspicion.
• Microcopy. Error messages, field labels, shipping information. It needs to sound like a native speaker wrote it, not like machine translation output.
• Prices in local currency, clear VAT and shipping info. Surprise costs at checkout are a guaranteed abandonment trigger.

Professional checkout and purchase flow localization is a separate workstream. It's not an add-on to product description translation. It's the foundation of conversion in a new market.

Emails and post-purchase communication

The customer bought. They get an order confirmation in English even though the entire store was in Turkish. A shipping email with dates in American format. A review request in a tone that feels rude in their culture.

These aren't hypothetical problems. They're real situations from localization projects that all end the same way: the customer who should have come back doesn't. Support tickets pile up, NPS drops.

Transactional emails and post-purchase communication must include:

• Every email in the customer's language. Order confirmation, tracking, invoice, review request, return notification.
• Local formats. A German expects DD.MM.YYYY and a comma as decimal separator. An American expects MM/DD/YYYY and a period. Small details, but they create the feeling that "this is my store."
• The right level of formality. Germans expect "Sie" (formal). French e-commerce often uses "tu" (informal). Wrong register immediately signals the brand doesn't know the market.

Trust signals. The customer doesn't know your brand

In a new market, nobody knows your brand. The customer hasn't seen your ads, hasn't gotten a recommendation from a friend, hasn't read about you in local media. They land on your site for the first time and decide in seconds: buy or leave.

What builds trust in a new market:

• Terms and return policy in the local language, compliant with local law. GDPR is the European standard, but implementation details vary between countries. Germany has one of the highest return rates in Europe. An unclear return policy is a dealbreaker.
• Reviews in the customer's language. Translated reviews carry far less weight than native ones. Trustpilot in Germany and the UK, Avis Vérifiés in France.
• Local trust badges. Trusted Shops in Germany, Keurmerk in the Netherlands. Seals the local customer actually recognizes.
• Phone number in local format and business hours in the customer's time zone.

AI as a new customer acquisition channel

A year ago this was conference talk. Now it's a real sales channel. More and more users start their buying journey in ChatGPT, Gemini, or Perplexity instead of Google. They're looking for comparisons, recommendations, answers to "is it worth it."

AI traffic converts differently from organic. Users arrive closer to a purchase decision because AI has already compared options for them. That means higher conversion and often a higher average basket.

What to do so AI cites your site:

• Decision-making content. Comparisons, rankings, "X vs Y" answers, "is it worth it" content. AI cites content that helps users decide.
• Schema markup. FAQ, Product, Review. Structured data makes it easier for AI to parse and cite your content.
• Brand presence in external sources. Industry articles, guest posts, directories. AI combines information from multiple sources. The more credible mentions, the higher your site appears in AI-generated answers.

This applies to each language version separately. It's not enough for your English site to be visible in AI. In the German market, you need German-language content that AI serving that market can find and cite.

Measure per market, not in aggregate

Everything above loses its value if you measure results as a single aggregate across the entire site.

5 million sessions sounds great. But if you don't know that the Czech market generates more revenue per session than Poland, or that Lithuania is growing fastest and deserves a bigger content budget, you're making decisions blind.

The minimum you need:

• GA4 with segments for each language version. Sessions, conversions, revenue. Separately for each market.
• Google Search Console: separate verification for each version. Otherwise you can't see which keywords are working in which market.
• Purchase path tracking per market. Where traffic comes from, how it converts, what the average order value is.
• Quarterly analysis, market by market. What's ranking, what's not. Where content gaps exist relative to local competition. Which market is growing, which needs intervention.

Checklist before entering a new market

1. URL structure. Subdirectory or subdomain, consistently. Separate sitemap and GSC verification per version.
2. Hreflang. Reciprocal tags, x-default, no conflict with canonicals.
3. Keywords. Separate research per market, not translated source-language keywords.
4. Content. Separate content strategy per market, not a copy-paste from the original version.
5. Platform. Does it give you control over checkout, hreflang, and emails?
6. Checkout. Local payment methods, address format, microcopy, currency with VAT.
7. Emails. Confirmations, tracking, invoices in the customer's language. Formats and formality.
8. Trust. Terms and returns compliant with local law. Reviews, trust badges, contact in local format.
9. AI. Decision-making content, schema markup, brand presence in external sources.
10. Measurement. GA4 and GSC per market. Purchase path tracking. Quarterly analysis.

None of these elements work in isolation. A fast store without localized checkout won't convert. Great content without correct hreflang won't rank. A good checkout without trust signals won't convince a customer seeing your brand for the first time.

Localization isn't a one-time project. It's systematically building presence in each market individually. But companies that take it seriously see results faster than those that "just translate the site."

Article by the team at tlumaczenie-stron-internetowych.pl, specializing in website and online store localization and multilingual SEO.